The devil is in the details: landscape features are insufficient to explain distribution patterns of non-native fishes in north Patagonian streams

Abstract

Ecological communities are structured by a combination of factors known as habitat templates. These templates work as a ´filter’ that permits some species with particular traits or phenotypes to establish and persist while excluding others. Defining which habitat variables and spatial scales have the stronger influence on freshwater communities is key to effective and efficient management of fluvial ecosystems. We took advantage of the relatively recent and well-studied history of salmonid introductions to Patagonia to evaluate if patterns of association of non-native species with abiotic factors vary between different spatial scales of the environmental filter. We characterized environmental variables at the basin and reach scale to assess their influence on the presence, abundance and structure of the salmonid assemblages in breeding streams. We found no evidence supporting that presence/absence patterns of salmonid distribution were strongly driven by landscape variables, except for basins with physical barriers to migration. However, we did find support for some climatic and geomorphological variables (e.g., precipitation and relief) influencing relative abundances. Our results do not support a scenario in which the distribution of any of the species affects the distribution of the other species, and suggest interference has played only a minor role in determining current fish distribution in the region. Instead, current patterns of presence and abundance of salmonids are best explained as the product of environmental filters. Our findings contribute to our understanding of the ecology of individual species and provide insight into mechanisms structuring fish assemblages in Southern Hemisphere’s lotic systems.

Introduction

Ecological communities are structured by combinations of factors defined as habitat templates (Southwood, 1977). These habitat templates mold evolutionary forces and structure ecological strategies for each of the species within the community. In other words, both the presence of a species at a particular site and its life history strategy are the result of environmental filtering (i.e., the tolerance of the species to a particular subset of biotic and abiotic characteristics) and trade-offs met during habitat adaptation.

Fluvial ecosystems are characterized by a physical habitat strongly influenced by the inherent hierarchical structure and patchiness that determine the distribution of organisms, food availability, predation and competition (Frissell et al., 1986, Schlosser & Kallemeyn, 2000). Fine-scaled variability results from the interaction between large-scale landscape variables (e.g., basin area, slope profile of stream-associated valleys and other geomorphological traits) and smaller scale variables (e.g., local structure and condition) (Frissell et al. 1986; Schlosser & Kallemeyn 2000). Because of that, defining which habitat variables and spatial scales have the most influence on freshwater communities is key to effective and efficient management of fluvial ecosystems (Matthews, 1998, Gido et al., 2006).

Environmental filtering is considered a major structuring mechanism of communities (Weiher & Keddy 1995). It is dominated by three ecological factors: dispersal restriction, abiotic environment and biotic interactions (Belyea & Lancaster, 1999). The first two act on a regional scale and delimit the area of action of the third, which operates on a local scale (Booth & Swanton, 2006). While the utility of the environmental filtering concept has been argued on the basis that it predicts patterns that cannot be distinguished from those produced by other mechanisms, such as competition (Kraft et al., 2015), there are good reasons to explore how patterns of trait or phylogenetic dispersion change in response to the environment (Cadotte & Tucker, 2017) .Therefore, if the relative effects of these general classes of factors in streams could be disentangled, then we would gain a comprehensive view of how each factor drives community composition.

In Patagonia, particularly at the scale of large drainage basins, fish distribution has been strongly influenced by the Andean uplift and the Quaternary glacial cycles (Hubert & Renno, 2006). After the retreat of glaciers during the Pleistocene, the ability of Patagonian fish to colonize post-glacial water bodies determined their present distribution, clearly constrained by climate, especially by temperature (Cussac et al., 2004, Ruzzante et al., 2006). In recent times, native freshwater fish communities have been altered on repeated occasions by the introduction of non-native species to generate sport fisheries, resulting in communities where up to six native species and four introduced salmonid species coexist within the watershed (Macchi et al., 1999). Many of these species have stable populations at several large lakes that have been intensively studied (Cussac et al., 2014). In contrast, relatively little is known about fish assemblages in streams or about the environmental filters that structure riverine communities (Aigo et al., 2008, Pascual et al., 2002, 2007, Barriga et al.2013, Lallement et al. 2020).

Of the four species introduced since 1904, rainbow trout (Onchorhynchus mykiss, Walbaum 1792), brown trout (Salmo trutta, Linnaeus 1758) and brook trout (Salvelinus fontinalis, Mitchill 1814) are currently the most widely distributed and abundant species. The rapid adaptation of these salmonid species to the new environment enabled them to establish self-sustaining populations, obviating the need for continuous import of eggs and ensuring of a steady supply of fish from local reproducers. The short and well-known history of introductions in the region shows that these species have high dispersive capabilities and found practically no biological resistance to invasion (Pascual et al. 2007). Thus, it can be posited that their current distribution in Patagonia is the result of environmental filtering at different scales. Although the influence of the landscape scale factors has been evaluated for the two most abundant species of salmonids in previous works (Quirós, 1991, Aigo, et al., 2008; Lallement et al. 2020), it is not known whether these factors alone explain the richness of non-native species in the watershed.

The objective of this work was to examine watershed and reach-level patterns of introduced salmonid species in North Patagonian rivers and their association with watershed characteristics derived from remote sensing and topography data across an environmental gradient. We compared the relative variation in salmonid density across different kind of watersheds expecting that some environmental characteristics at a landscape and reach scale conditioned i) the presence of salmonids, ii) the abundance of each species, iii) the assemblage conformation and iv) the dominance of a species at a regional level. We characterized environmental variables at the basin and reach scale to assess their influence on the presence, abundance and structure of the salmonid assemblages in breeding streams of the Upper Limay river basin. Our findings contribute to our understanding of the ecology of individual species and provide insight into mechanisms structuring fish assemblages in Southern Hemisphere’s stream systems. Such information not only contributes to knowledge of the ecological processes structuring aquatic communities, but also helps direct conservation and management activities.

Methods

Study Area

The Upper Limay River basin is located between the provinces of Río Negro and Neuquén (~40°63’ S, 71° 70’ W), Argentina, and drains an area of 6.980 km2, most of it within the boundaries of Nahuel Huapi National Park (Figure 1). Originating in the eastern slopes of the Andes mountain range, the basin presents a complex hydrological network, with many streams, rivers and lentic water bodies. Due to rain-shadowing effects by the Andes, the area experiences a steep longitudinal climatic gradient going from 3000 mm of yearly precipitation in the West to less than 600 mm to the East in about 60 km (Paruelo et al., 1998). This climatic gradient results in an eastward transition from a cold-temperate rainforest to shrubby dry steppes. The watershed has a highly connected, complex hydrologic network characterized by deep oligotrophic lakes of varying size, interconnected by streams, ponds and wetlands. The main hub of the network is Nahuel Huapi Lake, with an area of 529 km2 and a maximum depth of 464 m, which collects most waters from the basin, and drains through the Limay River towards the Atlantic Ocean (Figure 1).

Figure 1

Selection of watersheds

Thirty-five basins fully representing the environmental gradient of the watershed were selected for this study (Figure 1). For each basin, one or more sections of 2nd order or higher were selected; selection criteria were size, registered physiographic changes, existence of natural or anthropic barriers and the accessibility to each section (Bain & Stevenson, 1999). Based on these criteria, some basins were represented by a single section, while others had two or more representative sections. On larger basins with enough altitudinal range (e.g., Machete, Gutiérrez, Ñireco, Ñirihauau and Chacabuco), additional sections on tributary streams were sampled.

Watershed variables

A total of 32 watershed attributes were chosen following the available literature (Angermeier & Winston, 1999, Oakes et al., 2005, Smith & Kraft, 2005), and were grouped into four distinct categories (Geomorphological, Climatological, Land Use and Vegetation) based on the general aspect of the environment measured by each variable (Table A-Appendix section). All categories were calculated using GIS tools (version 2.6) under the QGIS environment (QGIS Development Team, 2014), or using specific formulas following Bain and Stevenson (1999). Land Use and Vegetation data were obtained from existing digital map layers available from the National Geographic Institute, the National Institute of Agricultural Technology (INTA) and National Parks Administration’s Biodiversity Information System; these maps were complemented with satellite imagery from Google Earth. Watershed morphological variables were calculated based on a digital elevation model with a resolution of 30 m (Landsat 6TM+). Average annual precipitation (mm) and temperature were calculated based on a map interpolating averages of daily measures from 25 meteorological stations located within from 38°46’0S–41°30’0S and 70°03’0W–71°45’0W (Barros et al., 1983). The Normalized Differential Vegetation Index (NDVI) was calculated from Landsat 6TM+ satellite images during the summer season of 2014. Geoprocessing and zonal statistics were computed using Quantum Gis (QGIS) (version 2.6) and digital map data. All the variables are available in the Appendix section (Table A).

#TABLE 1 - Basin-level variables
basins <- read_csv("basins.csv")

## Rows: 35 Columns: 33

## -- Column specification --------------------------------------------------------
## Delimiter: ","
## chr  (1): Basin
## dbl (32): lat, lon, area, perimeter, length, drain_net, drain_area, shape, K...

## 
## i Use `spec()` to retrieve the full column specification for this data.
## i Specify the column types or set `show_col_types = FALSE` to quiet this message.

gt(basins)

## Warning in grepl("\n", x, fixed = TRUE): input string 1 is invalid UTF-8

`

Basin	lat	lon	area	perimeter	length	drain_net	drain_area	shape	Kc	relief	Zmean	Zmax	Zmin	BRR	temp	NDVI	precip	rock	forest	wood	summit	wetland	urban	coihue	lenga	steppe	wetland2	highmnt	TAI	TAM	TAR	Total
Acantuco	-40.6870	-71.8259	18.76	21.67	6.64	8.80	0.47	0.0106	1.40	1165	1286.18	1934	769	0.17545181	12.188889	0.22	2698.29	0.00	70.76	0.00	28.56	0.000	0.00	27.56	43.02	0.00	0.00	28.62	1.63	0.40	0.04	2.07
Blanco	-40.9863	-71.7269	14.52	17.75	4.72	4.72	0.32	0.0146	1.30	1241	1410.95	2014	773	0.26292373	9.433333	0.11	2386.56	0.00	43.92	0.00	54.10	0.000	0.00	13.71	40.63	0.00	0.00	45.66	0.00	0.00	0.00	0.00
Blest	-41.0244	-71.8452	12.50	20.48	6.80	10.39	0.83	0.0180	1.62	1031	1227.72	1797	766	0.15161765	13.500000	0.26	3113.83	0.00	77.32	0.00	17.29	0.000	0.00	14.08	62.48	0.00	0.00	0.00	9.50	1.60	0.32	11.43
Bonito	-40.7357	-71.5788	56.72	39.23	13.24	22.74	0.40	0.0023	1.46	1147	1384.99	1916	769	0.08663142	10.427778	0.22	2001.43	0.00	73.58	0.00	26.42	0.000	0.00	20.38	38.10	0.00	0.00	28.56	14.88	0.56	0.28	15.73
Bravo	-40.9687	-71.8035	46.91	38.84	10.38	12.61	0.27	0.0025	1.59	1132	1263.73	1901	769	0.10905588	11.994444	0.12	2775.10	0.00	68.04	0.00	31.95	0.000	0.00	0.00	30.31	0.00	0.00	35.96	0.00	0.00	0.00	0.00
Casa de Piedra	-41.1604	-71.5157	64.78	44.05	19.77	34.13	0.53	0.0013	1.53	1457	1473.44	2206	749	0.07369752	5.550000	0.07	2324.05	0.00	58.11	10.50	30.75	0.000	0.00	0.05	35.86	0.00	0.00	37.17	11.27	0.00	0.17	11.44
Cascada	-41.1561	-71.4530	12.53	20.42	7.67	7.67	0.61	0.0104	1.62	1355	1256.49	2149	794	0.17666232	16.538889	0.07	1926.90	0.00	54.93	20.67	24.37	0.000	0.00	0.00	17.56	0.00	0.00	20.43	39.70	0.00	0.00	39.70
Castilla	-41.0226	-71.3418	25.76	31.00	10.27	21.49	0.83	0.0079	1.71	654	925.18	1426	772	0.06368062	23.977778	0.11	1439.01	0.00	52.79	47.02	0.00	0.209	0.00	0.00	8.85	31.37	0.00	0.00	10.01	36.48	7.72	54.21
Chacabuco	-40.9954	-71.2287	134.94	70.89	25.73	68.23	0.51	0.0008	1.71	1189	1054.93	1954	765	0.04621065	18.483333	-0.02	1196.57	19.55	33.13	22.48	0.00	17.450	0.00	0.00	13.35	71.03	11.17	4.45	3.91	13.78	0.00	17.69
Challhuaco	-41.2366	-71.3091	41.62	28.95	9.56	12.92	0.31	0.0034	1.26	1289	1450.07	2228	939	0.13483264	7.405556	0.06	1834.49	20.04	54.40	21.40	4.16	0.000	0.00	0.00	51.83	0.00	0.00	31.43	42.43	0.00	2.86	45.29
Coluco	-40.9123	-71.6688	25.15	25.93	9.26	9.26	0.37	0.0043	1.45	1182	1438.06	1953	771	0.12764579	11.100000	0.13	2541.63	0.00	60.19	0.00	39.50	0.000	0.00	21.11	38.93	0.00	0.00	39.72	3.93	2.02	0.12	6.07
De la Quebrada	-41.3616	-71.2715	14.31	18.53	5.22	5.22	0.36	0.0134	1.37	969	1635.22	2103	1134	0.18563218	2.394444	0.04	1526.00	39.39	59.59	0.00	0.00	1.000	0.00	0.00	48.36	0.00	0.00	45.56	50.00	2.14	0.43	52.56
Del Medio	-41.1808	-71.2129	108.07	60.25	24.52	35.20	0.33	0.0005	1.62	935	979.40	1727	792	0.03813214	20.066667	-0.05	1248.79	0.00	0.00	25.64	1.91	1.000	0.00	0.00	7.07	72.01	4.52	1.41	20.56	5.76	0.00	26.32
Estacada	-40.7830	-71.5257	49.06	35.55	13.87	27.06	0.55	0.0029	1.42	1205	1447.21	1976	771	0.08687815	9.233333	0.18	1978.71	0.00	67.66	0.00	32.33	7.820	0.10	22.85	41.70	0.00	0.00	35.45	1.88	0.00	0.00	1.88
Frey	-41.1712	-71.7300	36.59	26.23	7.49	9.84	0.27	0.0048	1.21	1391	1363.49	2159	768	0.18571429	9.700000	0.20	2626.64	0.00	74.71	0.00	21.06	0.000	0.00	24.76	53.79	0.00	0.00	15.63	1.76	1.63	0.26	3.78
Gallardo	-40.8701	-71.8212	97.32	52.24	17.38	32.01	0.33	0.0011	1.48	1160	1273.73	1942	782	0.06674338	10.450000	0.10	2962.76	0.00	63.91	0.00	26.78	0.704	0.00	17.70	45.91	0.00	0.00	27.51	1.75	5.50	0.00	7.25
Gutierrez	-41.2067	-71.4326	160.17	63.17	28.73	60.57	0.38	0.0005	1.40	1614	1261.34	2383	769	0.05617821	11.177778	0.03	1741.80	0.00	60.22	10.85	17.55	0.000	0.00	6.07	25.99	0.00	0.00	20.90	1.05	7.64	0.00	8.69
Huemul	-40.8569	-71.4419	54.14	36.52	12.80	34.12	0.63	0.0038	1.39	1371	1416.06	2140	769	0.10710938	7.027778	0.12	2112.66	0.00	64.29	0.00	35.72	0.000	0.00	26.21	29.63	0.00	0.00	40.49	7.00	0.00	0.00	7.00
Las Minas	-41.2928	-71.1703	44.40	31.58	10.59	15.21	0.34	0.0031	1.33	563	1190.42	1503	940	0.05316336	16.705556	-0.04	1382.30	53.93	17.67	28.10	0.00	0.310	0.00	0.00	6.58	93.42	0.00	0.00	18.86	0.64	0.00	19.50
LLuvuco	-41.1457	-71.6111	27.19	24.98	6.40	9.01	0.33	0.0081	1.34	1389	1537.49	2190	801	0.21703125	4.922222	0.04	2507.22	0.00	71.01	0.00	27.95	0.000	0.00	2.98	51.82	0.00	0.00	44.21	1.34	0.11	0.00	1.45
Machete	-40.8373	-71.8332	193.67	72.96	26.40	88.88	0.46	0.0007	1.47	1190	1279.72	1945	755	0.04507576	11.605556	0.13	2942.88	0.00	63.42	0.00	28.57	3.230	0.00	19.26	43.64	0.00	0.00	32.51	4.01	4.48	0.24	8.73
Manzano-Jones	-40.9812	-71.2790	30.49	37.40	13.10	13.38	0.44	0.0026	1.90	1038	1174.19	1892	854	0.07923664	16.805556	0.10	1426.93	7.37	77.96	4.13	0.00	10.550	0.00	0.00	41.52	49.49	2.33	6.69	17.00	9.46	0.00	26.46
Millaqueo	-40.9743	-71.6598	52.00	42.31	16.96	35.64	0.69	0.0024	1.64	1252	1353.28	2018	766	0.07382076	11.116667	0.13	2310.23	0.00	63.32	0.00	36.67	0.000	0.00	16.13	30.13	0.00	0.00	35.83	11.01	0.00	0.35	11.36
Neuquenco	-40.5768	-71.6595	21.94	26.31	6.96	7.42	0.34	0.0070	1.57	1088	1128.61	1871	783	0.15632184	16.811111	0.30	2302.93	0.00	53.04	36.24	10.73	0.000	0.00	46.17	21.10	0.00	0.00	5.52	5.16	1.54	0.09	6.78
Newbery	-40.9801	-71.1867	27.83	29.27	6.56	12.97	0.47	0.0108	1.55	660	1066.44	1454	794	0.10060976	21.533333	-0.09	1137.50	60.25	0.00	19.38	0.00	19.840	0.00	0.00	0.00	100.00	0.00	0.00	0.00	0.00	0.00	0.00
搼㸱ireco	-41.2078	-71.3215	113.16	62.83	19.74	37.55	0.33	0.0009	1.65	1455	1292.83	2228	773	0.07370821	10.622222	0.01	1657.58	11.90	37.37	38.55	8.73	0.000	1.58	0.00	30.10	5.34	0.00	26.90	18.80	0.09	0.00	18.89
搼㸱irihuau	-41.2253	-71.1863	723.80	193.38	58.07	278.58	0.38	0.0001	2.01	1469	1191.72	2234	765	0.02529705	13.400000	-0.04	1203.16	41.89	18.99	9.30	3.60	8.940	0.02	0.00	16.92	60.04	5.48	13.08	8.92	0.54	0.00	9.47
Patiruco	-41.0652	-71.7491	24.43	24.84	6.53	13.00	0.53	0.0125	1.41	1224	1271.24	2001	777	0.18744257	13.511111	0.17	2165.42	0.00	80.76	0.00	18.38	0.000	0.00	23.82	58.37	0.00	0.00	17.81	12.13	0.00	2.55	14.69
Pedregoso	-40.9034	-71.3690	20.53	25.35	7.22	8.52	0.41	0.0080	1.57	1415	1567.94	2186	771	0.19598338	3.977778	0.00	1796.36	0.00	45.62	7.32	47.06	0.000	0.00	0.00	2.19	0.00	0.00	85.24	1.45	0.00	0.00	1.45
Pireco	-40.7283	-71.8834	125.48	61.19	21.53	59.93	0.48	0.0010	1.53	1183	1258.26	1937	754	0.05494659	12.344444	0.19	2800.08	0.00	65.65	0.00	21.35	3.830	0.00	26.24	41.35	0.00	0.00	28.80	0.76	0.69	0.00	1.45
Quintriqueuco	-40.9248	-71.3206	15.26	20.80	5.47	5.48	0.36	0.0120	1.49	1147	1526.57	1932	785	0.20968921	6.644444	0.12	1576.86	23.17	68.78	5.04	3.03	0.000	0.00	0.00	55.37	0.00	0.00	39.06	8.19	0.00	0.00	8.19
Ragintuco	-40.8126	-71.4787	38.76	34.24	11.62	22.04	0.57	0.0042	1.54	1304	1460.31	2075	771	0.11222031	8.116667	0.15	2132.30	0.00	57.67	0.00	42.34	0.000	0.00	14.40	41.51	0.00	0.00	37.18	1.10	0.00	0.00	1.10
Torrontegui	-41.2788	-71.4390	17.47	20.03	6.16	6.16	0.35	0.0093	1.34	1342	1622.19	2145	803	0.21785714	2.583333	0.03	1462.27	0.00	61.19	0.00	38.27	0.000	0.00	0.00	39.38	0.00	0.00	47.17	22.61	1.19	0.59	24.39
Tristeza	-41.2894	-71.3209	41.29	32.45	12.16	15.88	0.38	0.0026	1.41	1164	1648.61	2234	1070	0.09572368	3.455556	-0.01	1733.11	32.36	63.47	4.19	0.00	0.000	0.00	0.00	37.54	0.48	0.00	52.97	20.56	0.32	0.32	21.19
Uhueco	-41.1671	-71.6557	5.44	10.82	3.42	3.42	0.63	0.0537	1.30	1212	1530.51	2024	812	0.35438597	6.305556	0.15	2586.98	0.00	70.47	0.00	29.16	0.000	0.00	1.84	72.24	0.00	0.00	25.92	0.00	0.00	0.00	0.00

`{=html}

Local Variables

Physicochemical variables were collected in 42 reaches. Variables included geomorphological, substrate type, hydrological and water quality (Table B-Appendix section). Location and number of sections sampled by stream were determined by the pattern shape of channels, covered area and accessibility, which determined that streams with the largest area have a greater number of sections sampled. Thus, sections were selected based on particular characteristics of each stream such as changes in the relief or riverine vegetation, tributary unions and presence of ponds or waterfalls.

#TABLE 2 - Reach-level variables
reach <- read_tsv("reaches.tsv")

## Rows: 42 Columns: 36

## -- Column specification --------------------------------------------------------
## Delimiter: "\t"
## chr  (1): reach
## dbl (35): order, chanwidth, chandepth, bank_depth, bank_width, sinuosity, sl...

## 
## i Use `spec()` to retrieve the full column specification for this data.
## i Specify the column types or set `show_col_types = FALSE` to quiet this message.

gt(reach)

## Warning in grepl("\n", x, fixed = TRUE): input string 1 is invalid UTF-8

`

reach	order	chanwidth	chandepth	bank_depth	bank_width	sinuosity	slope	floodwidth	entrenchment	width_depth_r	lakekm	flowlength	valleylength	pools	wetwidth	silt	sand	gravel	cobble	boulder	bedrock	wood	velocity	discharge	turbidity	temp	cond	ph	tds	salt	benthos	TAI	TAM	TAR	Total
Acantuco	2	16.00	1.50	3.00	22.00	1.046320	0.055830144	22.00	1.3750000	10.666667	11.35	6.644130	6.35	0.20640569	8.00	0.00	70.00	10.000	10.00	10.000	2.00	9	0.24876923	303.24400	2.5066667	10.70	60.5	7.60	43.2	30.7	29.33	1.6313933	0.3968254	0.04409171	2.072310
Blest Arriba	2	12.40	1.45	2.90	19.00	1.186652	0.071875970	17.00	1.3709677	7.862069	0.65	6.799514	5.73	0.77000000	4.00	0.00	0.00	0.000	3.50	94.000	2.50	3	0.14428571	206.51750	0.7500000	10.60	33.0	7.73	23.5	18.2	750.15	13.8815442	0.1322052	0.13220518	14.145955
Bonito Toma de agua	2	15.00	1.40	2.80	15.00	1.342516	0.028977394	15.00	1.0000000	10.714286	2.40	13.237208	9.86	24.45528757	7.00	0.00	54.00	5.000	5.00	33.000	3.00	8	0.38671429	678.35000	0.4000000	7.30	57.7	8.05	40.9	26.2	108.16	12.9870130	0.7421150	0.55658627	14.285714
Bonito trampa	2	30.00	1.22	2.44	50.00	1.342516	0.037981725	50.00	1.6666667	24.590164	1.65	13.237208	9.86	0.00000000	8.00	0.00	60.00	0.000	20.00	20.000	0.00	18	0.19566667	589.24000	0.1000000	7.85	57.5	7.85	40.8	28.7	65.87	16.7910448	0.3731343	0.00000000	17.164179
Casa de Piedra Ab	3	30.70	0.80	1.60	39.00	1.198908	0.031989084	39.00	1.2703583	38.375000	3.19	19.770000	16.49	0.00000000	7.60	0.00	10.00	10.000	10.00	70.000	0.00	19	0.41033333	2742.38500	0.2200000	9.80	53.3	8.06	38.0	26.9	170.02	11.9241192	0.0000000	0.18066847	12.104788
Casa de Piedra Ar	3	19.00	0.75	1.50	27.00	1.198908	0.075854180	27.00	1.4210526	25.333333	3.78	19.770000	16.49	23.43000000	12.40	0.00	10.00	20.000	5.00	50.000	15.00	7	0.31846667	704.51000	0.2200000	9.80	53.3	8.06	38.0	26.9	72.22	10.6250000	0.0000000	0.15625000	10.781250
Castilla Abajo	4	5.76	0.93	1.86	6.17	1.202576	0.003999979	6.17	1.0711806	6.193548	0.14	10.270000	8.54	11.16279070	4.20	0.00	28.00	19.000	40.00	13.000	0.00	5	0.31057143	264.63250	1.5500000	12.40	89.3	8.04	63.3	44.5	1109.97	14.6179402	45.1827242	0.00000000	59.800664
Castilla Medio	2	4.26	0.93	1.86	78.00	1.202576	0.003999979	78.00	18.3098592	21.300000	6.73	10.270000	8.54	0.81061520	3.49	0.00	10.00	10.000	80.00	0.000	0.00	0	0.17008333	118.75437	1.3700000	13.60	127.4	7.60	90.7	63.1	1028.22	5.4027353	27.7854960	15.43638665	48.624618
Chacabuco Arriba	3	8.20	0.80	1.60	20.40	1.634708	0.001999997	20.40	2.4878049	10.250000	6.03	25.730000	21.80	20.12000000	3.00	20.00	10.00	60.000	10.00	0.000	0.00	0	0.43166667	282.77500	0.9300000	13.30	138.4	7.60	98.4	68.4	1432.33	0.7453416	11.1801242	0.00000000	11.925466
Chacabuco Des	3	7.00	0.50	1.00	50.00	1.180275	0.011999424	50.00	7.1428571	14.000000	0.10	17.050000	10.43	11.41031492	4.00	30.00	0.00	20.000	50.00	0.000	0.00	2	0.23933333	138.21750	0.8100000	17.30	134.9	7.60	96.2	68.9	1836.44	4.5641260	13.3663689	0.00000000	17.930495
Chacabuco Superior (ex manzano)	2	8.50	0.73	1.46	50.00	1.195238	0.015998635	50.00	5.8823529	12.142857	17.45	12.550000	10.50	0.00000000	2.50	85.00	0.00	15.000	0.00	0.000	0.00	11	0.04980000	16.77000	0.5900000	14.60	103.8	7.80	73.7	52.3	137.60	12.5000000	13.3333333	0.00000000	25.833333
Challhuaco Abajo	2	11.00	0.60	1.20	80.00	1.174447	0.031989084	80.00	11.4285714	18.333333	15.37	9.560000	8.14	0.00000000	8.00	5.00	15.00	35.000	35.00	10.000	0.00	0	0.34470588	491.32500	0.8100000	17.30	134.9	7.60	96.2	68.9	383.31	16.0714286	0.0000000	0.00000000	16.071429
Challhuaco Arriba	1	4.60	0.85	1.70	5.00	1.174447	0.051953207	5.00	1.0869565	5.411765	17.52	9.560000	8.14	21.82539683	2.10	5.00	5.00	40.000	40.00	10.000	0.00	9	0.39500000	254.45000	1.0200000	5.30	106.0	8.26	75.8	46.1	963.71	68.7830688	0.0000000	5.73192240	74.514991
De la Quebrada	1	7.80	0.80	1.60	7.80	1.115385	0.057935094	7.80	1.0000000	9.750000	45.86	5.220000	4.68	3.07692308	4.50	0.50	26.25	16.000	36.00	20.000	1.25	0	0.09122222	51.95250	0.1300000	9.20	134.1	8.05	95.1	63.8	531.68	50.0000000	2.1367521	0.42735043	52.564103
Del Medio	2	42.50	1.15	2.30	15.00	1.298729	0.005999928	15.00	0.3529412	36.956522	30.72	24.520000	18.88	42.76315789	2.90	0.00	60.00	8.000	30.00	2.000	0.00	1	0.00000000	0.00000	0.5000000	13.10	109.5	7.46	77.8	54.4	387.93	20.5592105	5.7565789	0.00000000	26.315789
Estacada	3	12.00	1.90	3.80	18.00	1.130399	0.051953207	16.50	1.3750000	5.947368	0.37	13.870000	12.27	0.09227683	6.80	0.00	10.00	0.000	30.00	60.000	0.00	5	0.31520000	721.44500	0.3200000	12.50	68.3	7.60	48.8	35.0	124.66	1.8750000	0.0000000	0.00000000	1.875000
Frey Abajo	2	14.00	1.50	3.00	20.00	1.044630	0.077842391	20.00	1.4285714	9.333333	0.19	7.490000	7.17	0.24315069	12.75	0.00	0.00	1.000	90.00	9.000	0.00	11	0.19516667	909.16000	1.3200000	16.80	19.3	7.58	13.7	13.7	131.94	1.7105263	1.1842105	0.26315789	3.157895
Frey Arriba	2	16.00	1.50	3.00	20.00	1.044630	0.099668652	20.00	1.2500000	10.666667	0.64	7.490000	7.17	0.00000000	12.00	0.00	0.00	10.000	0.00	0.000	90.00	0	0.19516667	909.16000	0.2600000	18.50	19.7	7.53	14.0	15.2	131.94	1.8115942	2.0703934	0.25879917	4.140787
Gallardo	2	31.00	1.60	3.20	31.00	1.203846	0.027992686	31.00	1.0000000	19.375000	1.48	18.780000	15.60	4.25000000	10.00	0.00	12.50	5.000	55.00	0.000	27.50	5	0.33020000	609.49500	0.1000000	14.20	20.9	8.80	15.0	14.2	564.00	1.7500000	5.5000000	0.00000000	7.250000
Gutierrez municipalidad (ruta)	3	14.00	1.20	2.40	100.00	1.166378	0.005999928	100.00	7.1428571	11.666667	0.41	6.730000	5.77	0.00000000	10.46	0.00	0.00	70.000	30.00	0.000	0.00	1	0.60271429	1782.51500	0.0100000	10.40	74.0	7.87	52.7	36.6	306.69	3.0729580	10.4084060	0.00000000	13.481364
Gutierrez Usina (gendarmes)	3	16.00	1.30	2.60	100.00	1.166378	0.011999424	100.00	6.2500000	12.307692	1.06	6.730000	5.77	0.00000000	5.20	0.00	70.00	10.000	20.00	0.000	0.00	1	0.23145455	809.66000	0.4500000	12.00	75.7	7.73	53.8	38.1	137.49	0.3101737	6.6687345	0.00000000	6.978908
Gutierrez Virgen	3	67.00	1.80	3.60	100.00	1.166378	0.001999997	100.00	1.4925373	37.222222	4.35	6.730000	5.77	0.00000000	16.00	10.00	30.00	20.000	40.00	0.000	0.00	0	0.26930769	1730.35500	0.6300000	13.40	72.5	7.93	51.6	37.3	222.09	0.5208333	5.9895833	0.00000000	6.510417
Huemul pcpal	3	72.00	1.50	3.00	1000.00	1.123793	0.043971638	1000.00	13.8888889	48.000000	1.59	12.800000	11.39	1.96071429	5.60	0.00	20.00	10.000	10.00	60.000	0.00	9	0.34720000	344.15500	0.1766667	10.30	63.6	7.40	45.0	31.6	6.68	4.6428571	0.0000000	0.00000000	4.642857
Huemul secundario	3	72.00	1.50	3.00	1000.00	1.123793	0.043971638	1000.00	13.8888889	48.000000	1.59	12.800000	11.39	0.00000000	3.40	0.00	5.00	75.000	15.00	5.000	0.00	2	0.33785714	283.86750	0.4300000	10.00	63.2	7.20	44.9	31.4	6.68	9.3623890	0.0000000	0.00000000	9.362389
Jones	1	5.00	0.70	1.40	12.00	1.539197	0.003999979	12.00	2.4000000	1.539197	19.13	16.100000	10.46	12.83000000	3.10	0.00	50.00	50.000	0.00	0.000	0.00	2	0.12257143	29.11000	0.4800000	13.50	92.8	7.80	66.1	46.9	80.50	21.5053763	5.5913978	0.00000000	27.096774
Las Minas	2	7.00	0.55	1.30	20.43	1.127796	0.013999085	20.43	2.9185714	12.727273	37.57	10.590000	9.39	10.54000000	0.50	26.70	40.00	6.000	6.70	20.600	0.00	7	0.26550000	11.91600	1.4800000	16.20	258.0	8.73	184.0	128.0	8159.60	37.7289377	0.7326007	0.00000000	38.461538
Lluvuco	2	16.06	1.04	2.08	16.06	1.056106	0.111535184	21.80	1.3574097	11.471429	0.30	6.400000	6.06	15.12000000	17.27	0.00	10.00	18.000	42.00	30.000	0.00	14	0.18600000	545.03500	0.0600000	10.70	23.2	7.65	16.5	13.5	143.56	1.3419216	0.1073537	0.00000000	1.449275
Machete	3	50.00	1.30	2.60	51.00	1.877667	0.005999928	51.00	1.0200000	38.461538	7.77	26.400000	14.06	0.00000000	16.00	0.00	90.00	10.000	0.00	0.000	0.00	2	0.16275000	825.39000	0.1266667	12.10	48.8	8.40	34.7	24.8	564.00	4.0094340	4.4811321	0.23584906	8.726415
Millaqueo abajo (1)	3	2.70	0.72	1.14	45.00	1.493333	0.007999829	45.00	16.6666667	3.750000	0.89	16.920000	14.15	3.61757106	17.00	0.00	40.00	1.000	39.00	20.000	0.00	1	0.40952941	1902.57500	0.1900000	9.00	25.0	8.20	18.2	13.7	118.60	8.5538626	0.0000000	0.08910274	8.642965
Millaqueo arriba (2)	3	31.50	0.60	1.20	31.50	1.195760	0.007999829	31.50	1.0000000	52.500000	1.58	16.920000	14.15	0.00000000	10.00	0.00	5.00	0.000	25.00	70.000	0.00	2	0.34970000	1576.53500	0.1500000	9.60	25.8	8.02	18.4	14.6	281.44	13.4693878	0.0000000	0.61224490	14.081633
搼㸱ireco Alto	1	6.60	0.77	1.54	8.00	1.208818	0.087773892	8.00	1.2121212	8.571429	14.90	19.740000	16.33	14.00000000	3.00	3.75	6.25	20.625	43.75	25.625	0.00	0	0.16672414	883.92250	0.1700000	8.50	81.8	8.10	58.1	37.3	146.33	16.0968661	0.0000000	0.00000000	16.096866
搼㸱ireco arriba	2	8.80	0.45	0.90	40.00	1.208818	0.011999424	40.00	4.5454545	19.555556	9.07	19.740000	16.33	3.03000000	7.70	7.50	37.50	7.500	32.50	15.000	0.00	0	0.26750000	327.90000	0.4500000	19.30	93.0	8.44	66.1	47.0	312.18	16.9696970	0.0000000	0.00000000	16.969697
搼㸱ireco Urbano	3	15.00	1.25	2.50	175.00	1.208818	0.009999667	175.00	11.6666667	12.000000	1.46	19.740000	16.33	0.00000000	8.00	2.00	5.00	13.000	15.00	65.000	0.00	0	0.43625000	714.45000	0.3600000	10.50	152.7	8.03	108.0	73.6	1815.40	18.2065217	0.2717391	0.00000000	18.478261
搼㸱irihuau Aerop	4	75.00	2.00	4.00	150.00	1.192892	0.005999928	150.00	2.0000000	37.500000	19.31	58.070000	48.68	0.00000000	19.50	2.00	40.00	4.000	4.00	50.000	0.00	0	0.21104762	2335.55000	0.6200000	8.90	72.6	8.00	51.5	35.1	127.97	2.1281124	1.5322409	0.00000000	3.702916
搼㸱irihuau alto	3	23.00	1.90	3.80	300.00	1.192892	0.033986908	300.00	13.0434783	12.105263	45.78	58.070000	48.68	18.30997151	14.40	0.00	15.00	15.000	30.00	40.000	0.00	0	0.55075000	2660.19875	0.0400000	14.30	59.1	8.25	42.0	31.5	425.35	3.1250000	0.6250000	0.00000000	3.750000
搼㸱irihuau Superior Br1	3	23.60	1.50	3.00	160.00	1.192892	0.021996452	160.00	6.7796610	15.733333	44.14	58.070000	48.68	7.60803408	10.70	0.00	5.00	20.000	15.00	60.000	0.00	0	0.55075000	2660.19875	0.0900000	6.10	56.7	7.94	40.2	25.0	76.66	11.5642118	0.3043214	0.00000000	11.868533
搼㸱irihuau superior Br2	3	13.03	1.00	2.00	160.00	1.192892	0.021996452	160.00	12.2793553	13.030000	44.14	58.070000	48.68	0.00000000	4.00	0.00	5.00	15.000	30.00	50.000	0.00	0	0.15475000	98.96125	0.0200000	8.70	66.3	7.82	45.0	30.7	72.77	23.3870968	0.0000000	0.00000000	23.387097
搼㸱irihuau Vado	3	65.00	1.60	3.20	500.00	1.192892	0.009999667	500.00	7.6923077	40.625000	32.58	58.070000	48.68	7.59000000	10.00	0.00	1.00	4.000	20.00	35.000	40.00	0	0.73938095	2848.29000	2.1100000	7.40	70.5	7.90	50.0	31.8	337.40	4.4159544	0.2374169	0.00000000	4.653371
Patiruco	2	15.40	1.27	2.54	17.40	1.070492	0.051953207	17.40	1.1298701	11.846154	0.12	6.530000	6.10	44.69987229	5.80	0.00	10.00	20.000	20.00	60.000	0.00	1	0.04545454	220.50000	0.2800000	8.90	31.9	7.60	22.3	16.6	113.22	12.1328225	0.0000000	2.55427842	14.687101
Pireco	3	33.00	1.50	3.00	70.00	1.218449	0.009999667	70.00	2.1212121	22.000000	8.61	21.530000	17.67	3.95847751	16.00	0.00	40.00	0.000	50.00	10.000	0.00	16	0.63652941	3305.80000	1.3200000	11.80	43.3	7.73	30.7	23.2	18.40	0.7612457	0.6920415	0.00000000	1.453287
Ragintuco	2	15.00	1.30	2.60	15.00	1.136986	0.091822098	15.00	1.0000000	11.538462	0.49	11.620000	10.22	0.04884319	9.00	0.00	15.00	10.000	25.00	50.000	0.00	15	0.45227273	792.99500	0.8766667	10.80	52.0	7.20	37.7	27.1	79.09	1.0958904	0.0000000	0.00000000	1.095890
Tristeza	2	6.20	0.30	0.60	7.20	1.209559	0.019997334	7.20	1.1612903	24.000000	42.08	13.160000	10.88	0.00000000	6.00	0.00	5.00	10.000	70.00	25.000	0.00	6	0.28625000	467.11750	0.4600000	5.00	85.5	7.86	60.8	38.1	192.36	20.5566097	0.3162555	0.31625553	21.189121

`{=html}

Fish Capture

Different stream segments that presented a succession of pool-riffle-pool habitats were sampled during the austral summer (from December to March) of 2013-2014. The minimum sampled area depended on the size of the selected reach. In each section, presence-absence data were collected by three-pass electrofishing with a Smith-Root mod 12B equipment along 50-m reach. Relative abundance, expressed as catch per unit effort (CPUEN), was standardized based on actual length of each sweep to number of fish caught per 100 m2. Electrofishing was conducted from downstream to upstream following a zigzag trajectory and exploring all habitat types. The extremes were not blocked with nets. For basins sampled in more than one section, an average of the CPUEN of all fished sections was used as basin-level fish abundance. The Administration of Nahuel Huapi National Park and San Carlos de Bariloche Town Council approved our protocols and procedures and granted permission to collect fish samples (APN project n° 1173 and S.C de Bariloche Council note n° 412/SSMA/15).

Results

Introduced salmonid species are unevenly distributed among basins

Only two specimens of native species were caught: one of Galaxias maculatus (Jennyns, 1842) in the Frey basin and another of Hatcheria macraei (Girard, 1885) in the Ñirihuau basin. In contrast, a total of 4531 salmonids were caught. We found individuals aged between 0 and 3 years old at all sites, but older specimens (up to 8 years old) were caught only in some of the Eastern basins. No fish were captured in the Newbery, Blanco, Bravo and Uhueco basins; these streams present physical barriers (large waterfalls) restricting upstream movement of fish from the rest of the system (Figure 2).

Oncorhynchus mykiss was present in all basins where fish were caught. This species was also the most abundant in almost all basins (31/35); Salmo trutta, though present in 21/35 basins, was dominant in only 4 of them (Figure 2). Salvelinus fontinalis, was captured in 15/35 basins, but never dominated the assemblage (Figure 2).

Co-occurrence of introduced salmonid species varied among basins. The three species were found together in 10 basins. Only two out of three possible bispecific assemblages were found: rainbow trout with brown trout (S.trutta +O. mykiss, n = 10) and brown trout with brook trout (S.fontinalis + O.mykiss, n = 4). Monospecific assemblages were found only for rainbow trout (O. mykiss, n = 7).

No interspecific reciprocal influences on presence are evidenced by salmonid species distribution

Independence of species presence was tested using a contingency table approach, asking whether the probability of a species being present was contingent on the presence of other species.We calculated the expected frequency of each combination of species under the assumption that species distributions are independent of each other, based on the frequency each species was found. We then calculated the expected counts of all combinations, and compared them to the observed counts using a Fisher´s exact Test for contingency tables.

####Contingency test on joint species presence####

#Field data: number of basins in which a given species is present
Ntotal <- 35 #Total basins
NOm <- 31 #Basins in which rainbow trout Onchorhynchus mykiss (Om) is present
NSf <- 15 #Basins in which brook trout Salvelinus fontinalis (Sf) is present
NSt <- 21 #Basins in which brown trout Salmo trutta (St) is present

#Counts of basins in which each combination of species occur
NOb <- c(4,7,0,0,4,10,0,10)
names(NOb)<-c("none","Om","Sf","St","Om.Sf","Om.St","St.Sf","Om.Sf.St")

#Marginal probabilities
pOm <- NOm/Ntotal
pSf <- NSf/Ntotal
pSt <- NSt/Ntotal

#Calculate expected frequencies as joint probabilities
cOm <- pOm - pOm*pSf*pSt -(pOm*pSf - pOm*pSf*pSt) -(pOm*pSt - pOm*pSf*pSt)
cSf <- pSf - pOm*pSf*pSt -(pOm*pSf - pOm*pSf*pSt) -(pSt*pSf - pOm*pSf*pSt)
cSt <- pSt - pOm*pSf*pSt -(pOm*pSt - pOm*pSf*pSt) -(pSt*pSf - pOm*pSf*pSt)
cOm.Sf <- pOm*pSf - pOm*pSf*pSt
cOm.St <- pOm*pSt - pOm*pSf*pSt
cSt.Sf <- pSt*pSf - pOm*pSf*pSt
cOm.Sf.St <- pOm*pSf*pSt
none <- 1-sum(cOm,cSf,cSt,cOm.Sf,cOm.St,cSt.Sf,cOm.Sf.St)

#Build a vector of expected frequencies
frExp <- c(none,cOm,cSf,cSt,cOm.Sf,cOm.St,cSt.Sf,cOm.Sf.St)
names(frExp)<-c("none","Om","Sf","St","Om.Sf","Om.St","St.Sf","Om.Sf.St")

#Plot a Venn diagram
col.scheme <- c("red","green","blue")
draw.triple.venn(pOm,pSt,pSf,pOm*pSt,pSt*pSf,pSf*pOm,pOm*pSt*pSf, c("Om","St","Sf"), fill=col.scheme, cat.col=col.scheme, print.mode="percent")

## (polygon[GRID.polygon.11], polygon[GRID.polygon.12], polygon[GRID.polygon.13], polygon[GRID.polygon.14], polygon[GRID.polygon.15], polygon[GRID.polygon.16], text[GRID.text.17], text[GRID.text.18], text[GRID.text.19], text[GRID.text.20], text[GRID.text.21], text[GRID.text.22], text[GRID.text.23], text[GRID.text.24], text[GRID.text.25], text[GRID.text.26])

#Calculate vector of expected frequencies
NExp <- frExp*Ntotal

#Build a table of expected and observed occurrences.
expvsobs <- cbind(round(NExp),NOb)
colnames(expvsobs)<-c("Expected","Observed")
expvsobs <- as_tibble(expvsobs, rownames = "Species")
gt(expvsobs)

Species	Expected	Observed
none	1	4
Om	7	7
Sf	1	0
St	1	0
Om.Sf	5	4
Om.St	11	10
St.Sf	1	0
Om.Sf.St	8	10

#Performs Fisher's exact test for testing the null of independence of rows and columns in a contingency table with fixed marginals
ftest_all_basins <- fisher.test(expvsobs %>% select(Expected, Observed))
ftest_all_basins

## 
##  Fisher's Exact Test for Count Data
## 
## data:  expvsobs %>% select(Expected, Observed)
## p-value = 0.7537
## alternative hypothesis: two.sided

## Now repeat all of the above, but dropping those basins where no fish was caught.
## This explores that those basins are categorically different due to insurmountable obstacles for dispersal.
#Field data: number of basins in which a given species is present
Ntotal <- 31
NOm <- 31
NSf <- 15
NSt <- 21

#Counts of basins in which each combination of species occur
NOb <- c(7,0,0,4,10,0,10)
names(NOb)<-c("Om","Sf","St","Om.Sf","Om.St","St.Sf","Om.Sf.St")

#Marginal probabilities
pOm <- NOm/Ntotal
pSf <- NSf/Ntotal
pSt <- NSt/Ntotal

#Calculate expected frequencies as joint probabilities
cOm <- pOm - pOm*pSf*pSt -(pOm*pSf - pOm*pSf*pSt) -(pOm*pSt - pOm*pSf*pSt)
cSf <- pSf - pOm*pSf*pSt -(pOm*pSf - pOm*pSf*pSt) -(pSt*pSf - pOm*pSf*pSt)
cSt <- pSt - pOm*pSf*pSt -(pOm*pSt - pOm*pSf*pSt) -(pSt*pSf - pOm*pSf*pSt)
cOm.Sf <- pOm*pSf - pOm*pSf*pSt
cOm.St <- pOm*pSt - pOm*pSf*pSt
cSt.Sf <- pSt*pSf - pOm*pSf*pSt
cOm.Sf.St <- pOm*pSf*pSt
none <- 1-sum(cOm,cSf,cSt,cOm.Sf,cOm.St,cSt.Sf,cOm.Sf.St)

#Build a vector of expected frequencies
frExp <- c(cOm,cSf,cSt,cOm.Sf,cOm.St,cSt.Sf,cOm.Sf.St)
names(frExp)<-c("Om","Sf","St","Om.Sf","Om.St","St.Sf","Om.Sf.St")

#Plot a Venn diagram
col.scheme <- c("red","green","blue")
#draw.triple.venn(pOm,pSt,pSf,pOm*pSt,pSt*pSf,pSf*pOm,pOm*pSt*pSf, c("Om","St","Sf"), fill=col.scheme, cat.col=col.scheme, print.mode="percent")

#Calculate vector of expected frequencies
NExp <- frExp*Ntotal

#Build a table of expected and observed occurrences.

expvsobs.2 <- cbind(round(NExp),NOb) 
colnames(expvsobs.2)<-c("Expected","Observed")
expvsobs.2 <- as_tibble(expvsobs.2, rownames = "Species")
gt(expvsobs.2)

Species	Expected	Observed
Om	5	7
Sf	0	0
St	0	0
Om.Sf	5	4
Om.St	11	10
St.Sf	0	0
Om.Sf.St	10	10

#Performs Fisher's exact test for testing the null of independence of rows and columns in a contingency table with fixed marginals
ftest_no_empty_basins <- fisher.test(expvsobs.2 %>% select(Expected, Observed))
ftest_no_empty_basins

## 
##  Fisher's Exact Test for Count Data
## 
## data:  expvsobs.2 %>% select(Expected, Observed)
## p-value = 0.9378
## alternative hypothesis: two.sided

Fisher’s exact test failed to reject the hypothesis of independence (p = 0.753685), suggesting that species interactions are not needed to explain the frequency of each combination. Expected and observed counts are even more similar when basins without fish captures are excluded from the calculations (p = 0.9377599).

Salmonid abundances varied with environmental variables.

We then used generalized linear models (GLMs) on presence/absence and relative abundance data to assess the filtering influence of landscape and local variables. For each level and response variable type, we tried two approaches: 1) reducing the dimensionality of the independent variables through a principal components analysis (PCA), and then using the principal components explaining most variance as predictor variables in the GLM; and 2) by selecting independent variables that were not significantly correlated and using them as predictor variables in the GLM. While these two approaches are expected to yield qualitatively similar results, the second approach lends to more straightforward interpretations. All analyses were run using the function glm from the base package of the R computing environment R Core Team, version 3.5.1 (2018). To regress environmental variables using presence/absence as a response variable, we used a binomial logistic regression with a logit link. When the response variable was the logarithm of relative abundance, we used a Gaussian regression with an identity link.

#### Basin-level correlation analysis ####
basin.cpue.cor <- cor(basins[,-1], use="pairwise.complete.obs", method="spearman")[,18:21]
basin.cpue.cor.test <- rcorr(as.matrix(basins[,-1]), type="spearman") 
basin.cpue.cor.test$r[,29:32] %>%
  as_tibble(rownames = "variable")

## # A tibble: 32 x 5
##    variable       TAI     TAM    TAR   Total
##    <chr>        <dbl>   <dbl>  <dbl>   <dbl>
##  1 lat        -0.512  -0.109  -0.198 -0.538 
##  2 lon         0.434   0.0728 -0.182  0.475 
##  3 area       -0.0271  0.248  -0.221  0.0772
##  4 perimeter  -0.0273  0.304  -0.243  0.104 
##  5 length      0.0842  0.331  -0.207  0.213 
##  6 drain_net   0.0430  0.267  -0.131  0.178 
##  7 drain_area -0.0321 -0.141   0.104  0.0311
##  8 shape      -0.0847 -0.345   0.213 -0.191 
##  9 Kc          0.0955  0.187  -0.187  0.204 
## 10 relief     -0.124  -0.350  -0.129 -0.178 
## # ... with 22 more rows

Total catch of salmonids in streams from the upper Limay basin increased along a NW-SE gradient (Table 2). This gradient in relative abundance was associated to basins having a larger proportion of lowland, shrubby environments characterized by low rainfall, open woodlands and fewer rocky outcrops. This pattern in total catch was likely driven by dominance of rainbow trout, since its relative abundances presented the same pattern as the total catch (with the addition that abundance was higher at lower basins).

In contrast, brown trout abundances were not clearly associated with geographic gradients (Table 2). Instead, higher abundances were associated with lower, warmer and flatter basins, with fewer High-Andean habitat and more wetlands. Brook trout abundance showed no significant correlations with any environmental variable. However, spotty occurrence and low overall abundance of this species resulted in very low statistical power to detect any existing correlations.

Patterns of salmonid distribution are not strongly driven by landscape variables

Having ruled out a role for the interaction between species in their joint distribution (see above), we evaluated the influence of basin characteristics on individual species presence and abundances. We derived 28 watershed variables from remote sensing and topographical data sources.

#### Basin-level PCA and GLM analysis ####
basin.pca <- prcomp(basins[,2:29],center = TRUE, scale. = TRUE)
summary_basin.pca <- summary(basin.pca)
biplot(basin.pca)

basins$p.TAI <- ifelse(basins$TAI==0,0,1)
basins$p.TAR <- ifelse(basins$TAR==0,0,1)
basins$p.TAM <- ifelse(basins$TAM==0,0,1)

fviz_pca_ind(basin.pca, geom.ind = "point", 
             col.ind = "#FC4E07", 
             axes = c(1, 2), 
             pointsize = 1.5) 

fviz_screeplot(basin.pca, choice="eigenvalue")

basins <- cbind(basins, basin.pca$x[,1:3])

total.glm<-glm(Total~PC1+PC2+PC3, data=basins)
summary(total.glm)

## 
## Call:
## glm(formula = Total ~ PC1 + PC2 + PC3, data = basins)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -32.526   -6.982   -1.672    4.607   33.232  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  13.8917     2.2405   6.200 6.95e-07 ***
## PC1           1.3951     0.7351   1.898   0.0671 .  
## PC2          -2.1101     1.0755  -1.962   0.0588 .  
## PC3           2.1466     1.0898   1.970   0.0578 .  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 175.6907)
## 
##     Null deviance: 7437.3  on 34  degrees of freedom
## Residual deviance: 5446.4  on 31  degrees of freedom
## AIC: 285.98
## 
## Number of Fisher Scoring iterations: 2

p.tai.glm<-glm(p.TAI~PC1+PC2+PC3, data=basins, family="binomial")
summary(p.tai.glm)

## 
## Call:
## glm(formula = p.TAI ~ PC1 + PC2 + PC3, family = "binomial", data = basins)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -2.4523   0.1155   0.2870   0.4394   0.8824  
## 
## Coefficients:
##             Estimate Std. Error z value Pr(>|z|)   
## (Intercept)  3.13791    1.07382   2.922  0.00348 **
## PC1          0.43453    0.31319   1.387  0.16530   
## PC2          1.03896    0.56752   1.831  0.06715 . 
## PC3         -0.04685    0.24091  -0.194  0.84579   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 24.877  on 34  degrees of freedom
## Residual deviance: 19.421  on 31  degrees of freedom
## AIC: 27.421
## 
## Number of Fisher Scoring iterations: 7

tai.glm<-glm(TAI~PC1+PC2+PC3, data=basins)
summary(tai.glm)

## 
## Call:
## glm(formula = TAI ~ PC1 + PC2 + PC3, data = basins)
## 
## Deviance Residuals: 
##      Min        1Q    Median        3Q       Max  
## -19.3260   -6.5525    0.1208    5.3114   27.9262  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  10.6617     1.8193   5.860 1.83e-06 ***
## PC1           0.5367     0.5969   0.899  0.37549    
## PC2          -1.1018     0.8733  -1.262  0.21646    
## PC3           3.1618     0.8849   3.573  0.00118 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 115.8403)
## 
##     Null deviance: 5348.1  on 34  degrees of freedom
## Residual deviance: 3591.1  on 31  degrees of freedom
## AIC: 271.41
## 
## Number of Fisher Scoring iterations: 2

tai.glm<-glm(TAI~PC1+PC2+PC3, data=basins)
summary(tai.glm)

## 
## Call:
## glm(formula = TAI ~ PC1 + PC2 + PC3, data = basins)
## 
## Deviance Residuals: 
##      Min        1Q    Median        3Q       Max  
## -19.3260   -6.5525    0.1208    5.3114   27.9262  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  10.6617     1.8193   5.860 1.83e-06 ***
## PC1           0.5367     0.5969   0.899  0.37549    
## PC2          -1.1018     0.8733  -1.262  0.21646    
## PC3           3.1618     0.8849   3.573  0.00118 ** 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 115.8403)
## 
##     Null deviance: 5348.1  on 34  degrees of freedom
## Residual deviance: 3591.1  on 31  degrees of freedom
## AIC: 271.41
## 
## Number of Fisher Scoring iterations: 2

log.tai.glm<-glm(log(TAI+0.00001)~PC1+PC2+PC3, data=basins)
summary(log.tai.glm)

## 
## Call:
## glm(formula = log(TAI + 1e-05) ~ PC1 + PC2 + PC3, data = basins)
## 
## Deviance Residuals: 
##      Min        1Q    Median        3Q       Max  
## -11.6393   -0.2021    1.3702    2.5841    3.6290  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)
## (Intercept)   0.3574     0.7689   0.465    0.645
## PC1           0.1810     0.2523   0.717    0.478
## PC2           0.4053     0.3691   1.098    0.281
## PC3           0.1480     0.3740   0.396    0.695
## 
## (Dispersion parameter for gaussian family taken to be 20.69327)
## 
##     Null deviance: 680.34  on 34  degrees of freedom
## Residual deviance: 641.49  on 31  degrees of freedom
## AIC: 211.12
## 
## Number of Fisher Scoring iterations: 2

p.tam.glm<-glm(p.TAM~PC1+PC2+PC3, data=basins, family="binomial")
summary(p.tam.glm)

## 
## Call:
## glm(formula = p.TAM ~ PC1 + PC2 + PC3, family = "binomial", data = basins)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -1.7402  -1.1886   0.7392   1.0056   1.4048  
## 
## Coefficients:
##             Estimate Std. Error z value Pr(>|z|)
## (Intercept)   0.5059     0.3826   1.322    0.186
## PC1           0.2536     0.1640   1.546    0.122
## PC2           0.1758     0.2411   0.729    0.466
## PC3          -0.1049     0.1706  -0.615    0.539
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 47.111  on 34  degrees of freedom
## Residual deviance: 43.254  on 31  degrees of freedom
## AIC: 51.254
## 
## Number of Fisher Scoring iterations: 4

tam.glm<-glm(TAM~PC1+PC2+PC3, data=basins)
summary(tam.glm)

## 
## Call:
## glm(formula = TAM ~ PC1 + PC2 + PC3, data = basins)
## 
## Deviance Residuals: 
##      Min        1Q    Median        3Q       Max  
## -11.4313   -2.5794   -0.9743    1.8485   25.0488  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)   
## (Intercept)   2.7591     0.9938   2.776  0.00924 **
## PC1           0.8081     0.3261   2.478  0.01885 * 
## PC2          -0.7949     0.4771  -1.666  0.10575   
## PC3          -0.9194     0.4834  -1.902  0.06650 . 
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 34.57034)
## 
##     Null deviance: 1505.0  on 34  degrees of freedom
## Residual deviance: 1071.7  on 31  degrees of freedom
## AIC: 229.08
## 
## Number of Fisher Scoring iterations: 2

log.tam.glm<-glm(log(TAM+0.00001)~PC1+PC2+PC3, data=basins)
summary(log.tam.glm)

## 
## Call:
## glm(formula = log(TAM + 1e-05) ~ PC1 + PC2 + PC3, data = basins)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -10.422   -5.904    2.523    4.789    8.267  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  -4.3252     1.0037  -4.309 0.000154 ***
## PC1           0.6148     0.3293   1.867 0.071378 .  
## PC2           0.1230     0.4818   0.255 0.800167    
## PC3          -0.4896     0.4882  -1.003 0.323693    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 35.26049)
## 
##     Null deviance: 1253.7  on 34  degrees of freedom
## Residual deviance: 1093.1  on 31  degrees of freedom
## AIC: 229.77
## 
## Number of Fisher Scoring iterations: 2

p.tar.glm<-glm(p.TAR~PC1+PC2+PC3, data=basins, family="binomial")
summary(p.tar.glm)

## 
## Call:
## glm(formula = p.TAR ~ PC1 + PC2 + PC3, family = "binomial", data = basins)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -1.7296  -1.0883  -0.5299   1.1070   1.4891  
## 
## Coefficients:
##             Estimate Std. Error z value Pr(>|z|)  
## (Intercept) -0.42512    0.39038  -1.089   0.2762  
## PC1         -0.31573    0.17710  -1.783   0.0746 .
## PC2         -0.25317    0.25557  -0.991   0.3219  
## PC3         -0.08613    0.17121  -0.503   0.6149  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 47.804  on 34  degrees of freedom
## Residual deviance: 42.691  on 31  degrees of freedom
## AIC: 50.691
## 
## Number of Fisher Scoring iterations: 5

tar.glm<-glm(TAR~PC1+PC2+PC3, data=basins)
summary(tar.glm)

## 
## Call:
## glm(formula = TAR ~ PC1 + PC2 + PC3, data = basins)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -1.7754  -0.5585  -0.2577   0.1115   5.9613  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)  
## (Intercept)  0.46686    0.23242   2.009   0.0533 .
## PC1          0.05132    0.07625   0.673   0.5059  
## PC2         -0.21428    0.11156  -1.921   0.0640 .
## PC3         -0.09477    0.11305  -0.838   0.4083  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 1.89062)
## 
##     Null deviance: 67.769  on 34  degrees of freedom
## Residual deviance: 58.609  on 31  degrees of freedom
## AIC: 127.37
## 
## Number of Fisher Scoring iterations: 2

log.tar.glm<-glm(log(TAR+0.00001)~PC1+PC2+PC3, data=basins)
summary(log.tar.glm)

## 
## Call:
## glm(formula = log(TAR + 1e-05) ~ PC1 + PC2 + PC3, data = basins)
## 
## Deviance Residuals: 
##    Min      1Q  Median      3Q     Max  
## -7.005  -4.859  -2.265   5.253   8.996  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  -6.9855     0.9037  -7.730 1.02e-08 ***
## PC1          -0.4800     0.2965  -1.619    0.116    
## PC2          -0.3564     0.4338  -0.822    0.418    
## PC3          -0.1365     0.4396  -0.311    0.758    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 28.5864)
## 
##     Null deviance: 983.14  on 34  degrees of freedom
## Residual deviance: 886.18  on 31  degrees of freedom
## AIC: 222.43
## 
## Number of Fisher Scoring iterations: 2

We found that many of the variables were highly correlated; thus, we used a principal component analysis (PCA) to reduce the dimensionality of variable space. The first three principal components (PC) explained 65.651 % of the total variance (PC1 34.155%; PC2 15.956%, PC3 15.54%).

We used these three PCs to test if basin characteristics influenced the presence of each species by fitting a logistic regression using GLMs (see Methods), but found no significant influence of either PC on presence for any of the species. We then repeated the same analysis using relative abundance (individuals caught per unit effort; CPUEN) as a dependent variable instead. We found that rainbow trout abundance was significantly influenced by PC3 (p=0.001, loaded mainly by Latitude and Zmin), while brown trout abundance was instead, significantly influenced by PC1 (p=0.002, loaded mainly by Steppe and precipitations). Brook trout was not influenced by any PCs (Table 3). Repeating the analyses using the logarithm of relative abundance as response variable yielded no significant results for any species or PCs.

Hence, for all three species, we found that the probability of a species being present in a basin was not explained by any of the basin characteristics we used in this analysis. However, we detected evidence that the abundance of rainbow trout and brown trout (but not brook trout) is influenced by different combinations of basin traits. Thus, the three species differ in their response to landscape level filtering variables.

Local variables influence weakly but differentially the abundance of each species

To evaluate the influence of local characteristics on species presence we estimated 31 variables using local data records (see Table 3). Since several local-scale variables presented significant correlations, we used PCA to reduce the dimensionality of the variable space.

#### Reach-level PCA and GLM analysis ####
reach<-read.table("reaches.tsv", header=TRUE, sep="\t")
reach$p.TAI <- ifelse(reach$TAI==0,0,1)
reach$p.TAR <- ifelse(reach$TAR==0,0,1)
reach$p.TAM <- ifelse(reach$TAM==0,0,1)

reach.pca <- prcomp(na.omit(reach[,2:32]),center = TRUE, scale. = TRUE)
summary_reach.pca <- summary(reach.pca)
biplot(reach.pca)

fviz_pca_ind(reach.pca, geom.ind = "point", 
             col.ind = "#FC4E07", 
             axes = c(1, 2), 
             pointsize = 1.5) 

fviz_screeplot(reach.pca, choice="eigenvalue")

fviz_screeplot(reach.pca, choice="variance")

fviz_contrib(reach.pca, choice="var", axes=3)

reach <- cbind(reach, reach.pca$x[,1:3]) 

r.total.glm<-glm(Total~PC1+PC2+PC3, data=reach)
summary(r.total.glm)

## 
## Call:
## glm(formula = Total ~ PC1 + PC2 + PC3, data = reach)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -18.893   -6.282   -2.203    2.488   47.972  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  16.6030     2.0181   8.227 5.75e-10 ***
## PC1           3.9703     0.7882   5.037 1.18e-05 ***
## PC2           1.0609     0.9774   1.085    0.285    
## PC3          -0.2520     1.1941  -0.211    0.834    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 171.0609)
## 
##     Null deviance: 11049.8  on 41  degrees of freedom
## Residual deviance:  6500.3  on 38  degrees of freedom
## AIC: 340.95
## 
## Number of Fisher Scoring iterations: 2

r.p.tai.glm<-glm(p.TAI~PC1+PC2+PC3, data=reach, family="binomial")
summary(r.p.tai.glm)

## 
## Call:
## glm(formula = p.TAI ~ PC1 + PC2 + PC3, family = "binomial", data = reach)
## 
## Deviance Residuals: 
##       Min         1Q     Median         3Q        Max  
## 2.409e-06  2.409e-06  2.409e-06  2.409e-06  2.409e-06  
## 
## Coefficients:
##               Estimate Std. Error z value Pr(>|z|)
## (Intercept)  2.657e+01  5.495e+04       0        1
## PC1         -1.717e-16  2.146e+04       0        1
## PC2          2.082e-10  2.661e+04       0        1
## PC3         -6.497e-11  3.251e+04       0        1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 0.0000e+00  on 41  degrees of freedom
## Residual deviance: 2.4367e-10  on 38  degrees of freedom
## AIC: 8
## 
## Number of Fisher Scoring iterations: 25

r.tai.glm<-glm(TAI~PC1+PC2+PC3, data=reach)
summary(r.tai.glm)

## 
## Call:
## glm(formula = TAI ~ PC1 + PC2 + PC3, data = reach)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -21.076   -6.200   -1.187    3.473   49.760  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  11.9807     1.8374   6.520  1.1e-07 ***
## PC1           2.7704     0.7176   3.861 0.000426 ***
## PC2           0.6754     0.8899   0.759 0.452562    
## PC3          -0.7988     1.0872  -0.735 0.467007    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 141.7978)
## 
##     Null deviance: 7659.9  on 41  degrees of freedom
## Residual deviance: 5388.3  on 38  degrees of freedom
## AIC: 333.07
## 
## Number of Fisher Scoring iterations: 2

r.log.tai.glm<-glm(log(TAI+0.00001)~PC1+PC2+PC3, data=reach)
summary(r.log.tai.glm)

## 
## Call:
## glm(formula = log(TAI + 1e-05) ~ PC1 + PC2 + PC3, data = reach)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -3.0062  -0.7251   0.3885   0.8647   1.9106  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  1.82758    0.18479   9.890 4.64e-12 ***
## PC1          0.21254    0.07217   2.945  0.00549 ** 
## PC2          0.10634    0.08949   1.188  0.24208    
## PC3         -0.05795    0.10934  -0.530  0.59918    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 1.434127)
## 
##     Null deviance: 69.363  on 41  degrees of freedom
## Residual deviance: 54.497  on 38  degrees of freedom
## AIC: 140.13
## 
## Number of Fisher Scoring iterations: 2

r.p.tam.glm<-glm(p.TAM~PC1+PC2+PC3, data=reach, family="binomial")
summary(r.p.tam.glm)

## 
## Call:
## glm(formula = p.TAM ~ PC1 + PC2 + PC3, family = "binomial", data = reach)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -1.7358  -1.2681   0.7601   0.8739   1.2453  
## 
## Coefficients:
##             Estimate Std. Error z value Pr(>|z|)  
## (Intercept)  0.73545    0.34421   2.137   0.0326 *
## PC1          0.09223    0.14663   0.629   0.5293  
## PC2          0.11422    0.16897   0.676   0.4991  
## PC3         -0.30266    0.21264  -1.423   0.1546  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 53.467  on 41  degrees of freedom
## Residual deviance: 50.270  on 38  degrees of freedom
## AIC: 58.27
## 
## Number of Fisher Scoring iterations: 4

r.tam.glm<-glm(TAM~PC1+PC2+PC3, data=reach)
summary(r.tam.glm)

## 
## Call:
## glm(formula = TAM ~ PC1 + PC2 + PC3, data = reach)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -12.344   -3.251   -1.789    1.620   39.693  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)   
## (Intercept)   3.9785     1.2848   3.097  0.00367 **
## PC1           0.9957     0.5018   1.984  0.05447 . 
## PC2           0.3825     0.6222   0.615  0.54238   
## PC3           0.3544     0.7602   0.466  0.64378   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 69.3289)
## 
##     Null deviance: 2948.8  on 41  degrees of freedom
## Residual deviance: 2634.5  on 38  degrees of freedom
## AIC: 303.02
## 
## Number of Fisher Scoring iterations: 2

r.log.tam.glm<-glm(log(TAM+0.00001)~PC1+PC2+PC3, data=reach)
summary(r.log.tam.glm)

## 
## Call:
## glm(formula = log(TAM + 1e-05) ~ PC1 + PC2 + PC3, data = reach)
## 
## Deviance Residuals: 
##    Min      1Q  Median      3Q     Max  
## -9.702  -6.654   2.851   4.018   6.707  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  -3.4370     0.9120  -3.769 0.000557 ***
## PC1           0.3985     0.3562   1.119 0.270279    
## PC2           0.3307     0.4417   0.749 0.458589    
## PC3          -0.6849     0.5396  -1.269 0.212100    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 34.93458)
## 
##     Null deviance: 1447.1  on 41  degrees of freedom
## Residual deviance: 1327.5  on 38  degrees of freedom
## AIC: 274.23
## 
## Number of Fisher Scoring iterations: 2

r.p.tar.glm<-glm(p.TAR~PC1+PC2+PC3, data=reach, family="binomial")
summary(r.p.tar.glm)

## 
## Call:
## glm(formula = p.TAR ~ PC1 + PC2 + PC3, family = "binomial", data = reach)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -1.8308  -0.7637  -0.3839   0.9686   2.2445  
## 
## Coefficients:
##             Estimate Std. Error z value Pr(>|z|)   
## (Intercept)  -1.1141     0.5261  -2.118  0.03421 * 
## PC1           0.1992     0.2207   0.903  0.36670   
## PC2          -0.7662     0.2922  -2.622  0.00875 **
## PC3          -0.6633     0.4834  -1.372  0.17003   
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 54.748  on 41  degrees of freedom
## Residual deviance: 42.257  on 38  degrees of freedom
## AIC: 50.257
## 
## Number of Fisher Scoring iterations: 6

r.tar.glm<-glm(TAR~PC1+PC2+PC3, data=reach)
summary(r.tar.glm)

## 
## Call:
## glm(formula = TAR ~ PC1 + PC2 + PC3, data = reach)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -2.0393  -0.8905  -0.4877   0.1706  13.9396  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)
## (Intercept) 0.642742   0.392217   1.639    0.110
## PC1         0.204868   0.153183   1.337    0.189
## PC2         0.002313   0.189950   0.012    0.990
## PC3         0.193184   0.232069   0.832    0.410
## 
## (Dispersion parameter for gaussian family taken to be 6.461028)
## 
##     Null deviance: 261.55  on 41  degrees of freedom
## Residual deviance: 245.52  on 38  degrees of freedom
## AIC: 203.35
## 
## Number of Fisher Scoring iterations: 2

r.log.tar.glm<-glm(log(TAR+0.00001)~PC1+PC2+PC3, data=reach)
summary(r.log.tar.glm)

## 
## Call:
## glm(formula = log(TAR + 1e-05) ~ PC1 + PC2 + PC3, data = reach)
## 
## Deviance Residuals: 
##     Min       1Q   Median       3Q      Max  
## -7.8960  -4.0011  -0.9523   3.9675  11.6834  
## 
## Coefficients:
##             Estimate Std. Error t value Pr(>|t|)    
## (Intercept)  -7.7087     0.7528 -10.241 1.76e-12 ***
## PC1           0.1362     0.2940   0.463  0.64590    
## PC2          -1.0812     0.3646  -2.966  0.00519 ** 
## PC3          -0.2874     0.4454  -0.645  0.52270    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for gaussian family taken to be 23.7996)
## 
##     Null deviance: 1128.75  on 41  degrees of freedom
## Residual deviance:  904.38  on 38  degrees of freedom
## AIC: 258.11
## 
## Number of Fisher Scoring iterations: 2

The first three principal components (PC) explained 45.192 % of the total variance (PC1 21.664 %; PC2 14.089%, PC3 9.439%).

We tested whether local characteristics influenced the presence-absence of each species by fitting a logistic regression using Generalized Linear Models, and found no significant influence of either PCs on presence-absence for rainbow trout and brown trout, while PC2 (loaded mainly by water quality and slope variables) had a significant influence on presence of brook trout (z-value=-2.62, p-value=0.009). We then tested if reach variables influence relative abundances (either absolute values or their logarithms) and found a significant relation between PC1 (loaded mainly by channel morphology and water quality variables) and rainbow trout abundances (absolute – t=3.86, p=0.0004 and logarithms values –t=2.94, p=0.0055), but no significant influence on brown trout. We found significant influence of PC2 on the logarithm of brook trout abundance (t-value=-2.97, p=0.0052) but not on the absolute values. In summary, our regression analysis showed that the influence of local traits varies across all three salmonid species.

Discussion

One of the most pervasive concepts in the study of community assembly is the metaphor of the environmental filter, which refers to abiotic factors that prevent the establishment or persistence of species in a particular location. However, this concept has been criticized because the evidence used in many studies to assess environmental filtering is insufficient to distinguish filtering from the evolutionary outcome of historical biotic interactions (Kraft et al. 2015). In our work we took advantage of the relatively recent and well-studied history of salmonid introductions to evaluate if non-native species show different patterns of association with abiotic factors at different spatial scales of the filter.

Successful establishment of introduced salmonid species can depend on both biotic resistance (by the native community of organisms) and environmental resistance (habitat suitability), as well as chance events (Moyle & Light, 1996, Karr et al.,1987, Fausch,2007). The success of its dispersion has been well studied in several environments such as New Zealand (McDowall, 2003), Europe (Korsu, Huusko & Muotka, 2008) and Chile (Arismendi et al., 2014, 2019 , Habit, 2015). When salmonids were introduced to Patagonian waters in 1904 (Baigún & Quirós, 1985), they found almost no biotic resistance for the invasion and achieved a wide distribution throughout the region (Pascual et al. 2007). We indeed found widespread presence of salmonid species and almost total absence of native species in our study area. Thus, the only expected source of current biotic interference would be interspecific interaction between the salmonid species themselves. However, the results of our contingence analysis on species distribution does not support a scenario in which any of the species is affecting the distribution of the other species. This suggests that the streams in our study area have not reached carrying capacity for salmonids. Alternatively, salmonid species might be partitioning the riverine habitats to minimize niche overlap, as reported in other parts of the world (Bozek & Hubert, 1992, Reeves, Bisson & Dambacher, 1998, Fausch, 2008, Marchetti et al., 2011).

Interspecific interference among salmonid species has been proposed in previous studies in this region (Juncos, Beauchamp & Vigliano, 2013, Arismendi et al., 2014); we found however no evidence for it. Thus, we can assume that interference has played only a minor role in determining current fish distribution in the region. Instead, current patterns of presence and abundance of salmonids are best explained as the product of environmental filters. Indeed, reasons for the successful invasion reported from southern Chile has been related in part to the excellent abiotic conditions they found in the region (Pascual et al., 2005, Habit et al., 2012, Habit et al., 2015).

Biological invasions are inherently complex. A successful invader must survive a series of events: transport to the invasion site, initial establishment, spread to a broad area, and then integration into the existing biotic community (Moyle & Light, 1996, Kolar & Lodge, 2001). Not surprisingly, most introduced species fail to become established and reach invasive status (Moyle and Light 1996, Arismendi et al., 2014). Our results suggest that in North Patagonia, biotic resistance from native fish species seems to have had little or no influence on the invasion process by introduced salmonid species, and that success of invaders in the face of low odds is related, as we previously suggested, mostly with the presence of favorable environments, such as flow regime (conditioned for the rainfall gradient) and food availability (Lallement et al. 2020).

In contrast to other reports (Marchetti, Moyle & Levine, 2004, Stanfield, Scott & Borwick, 2006), we saw no evidence that presence/absence patterns of salmonid distribution were strongly driven by landscape variables, except for those basins with environmental barriers. However, when analyzing responses of relative abundances (CPUEN), the influence of climatic and geomorphological variables (e.g., precipitation and relief) became more evident. These two types of factors have been mentioned in other systems as determinants of salmonid distribution (Stanfield et al., 2006, Warren, Dunbar & Smith, 2015): rainfall and geology have a direct influence on stream discharge and are thus determinant during early development of salmonid life cycles (Heggenes & Traaen, 1988,Nehring & Anderson, 1993).

Landscape characteristics (e.g., general slope of the valley and geomorphological aspects) determine local river section traits such as substrate composition, pool dimensions and refuge availability, which in turn strongly correlate with the structure and distribution of the assemblages of fish in a basin (Fischer & Paukert, 2008,Rowe, Pierce & Wilton, 2009). For our study area, we found evidence that some local traits modulate rainbow trout abundance but do not explain abundances of brown trout and brook trout, or the presence-absence patterns for all three salmonid species. Low abundances for a species could be causing diminished statistical power to detect the influence of environmental variables; this could be affecting our results for brook trout. Differential abundances are nonetheless likely to result from differential responses to the same environment by each species. Thus, the layered influence of environmental filters was reflected in a weak but differential influence of local traits on abundance of each salmonid species.

Habitat suitability for salmonids is often controlled by local variables as temperature regime (Rahel & Nibbelink, 1999,Harig & Fausch, 2002, Coleman & Fausch, 2007), flow regime (Strange & Foin, 1999, Fausch et al., 2001), stream size (Rahel & Nibbelink, 1999) and habitat factors correlated with stream gradient and channel geomorphology (Fausch, 1989, Montgomery et al., 1999). Due to the steep precipitation gradient in our study area, several streams originate in wooded areas and cross wide valleys where shrubby vegetation typical of arid steppes predominates. Shrubby riverbanks result in higher solar irradiation and air temperatures, favoring higher primary production of the periphyton and sustaining an important biomass of macroinvertebrates (Miserendino, 2007, Modenutti et al., 2010). It is these streams where we found higher fish abundances.

A hierarchical perspective of stream systems, whereby properties at the site level are constrained by processes occurring in the catchment, provides a useful analytical framework (Vannote et al., 1980, Frissell et al., 1986, Imhol, Fitzgibbon & Annable, 1996). Previous predictive modeling studies have indicated that landscape-scale watershed characteristics are better to explain fish distributions than reach-scale characteristics (Creque, Rutherford & Zorn, 2005, Frimpong et al., 2005). However, this study did not find that watershed-scale variables were significantly better at explaining or predicting fish distributions in North Patagonia. The most likely explanation for this absence of significance can be the lack of habitat saturation by fishes due to temporal variability in fish distributions, unmeasured explanatory (reach- and watershed-scale characteristics) variables in the data set or insufficient sample size. As Stanfield and Gibson (2006), we do not contend that landscape variables are important for salmonids, but suggest that landscape traits condition the range of geomorphological variation inside a watershed that ultimately defines the current densities in a section as we previously observed (Lallement et al. 2020). Results from this study suggest further information is needed to understand how the variability of spatial scale affects the distribution of fishes in streams of Patagonia.

Conservation of stream habitats and their biota requires an understanding of how environmental factors, both natural and human-influenced, structure aquatic assemblages across different scales of space and time (Fausch, Baxter & Murakami, 2010,Grossman, Warnell & Sabo, 2010). Many previous studies have focused on the importance of variables at the local habitat scale or across relatively small study regions. Although not explored in this study, distributions of fishes are variable and may be regulated by seasonal movements and the frequency and duration of stochastic flow cycles. In any case, it is important to bear in mind that the results obtained in regions with different histories of colonization and introduction are not necessarily extrapolated. By including a comprehensive suite of variables at multiple spatial scales in a large region, our study contributes to understanding how these environmental factors may interact to structure invasive fish communities. Moreover, considering the pristine (or near-pristine) condition of the streams sampled here, the relationships observed between fishes and landscape variables can be used as a reference for further studies addressing the effects of human modifications on aquatic biodiversity of North Patagonia.

Acknowledgments

We thank to the National Park Administration (APN) that through the Biodiversity Information System (SIB) provided us with the data for Use of Land and Vegetation in watersheds. We thank all the members of the Grupo de Evaluación y Manejo de Recursos Icticos (GEMaRI) for their field assistance. Funding was provided by the Agencia Nacional de Promoción de Ciencia y Tecnología, Argentina (PICT projects 016 and 2959). The authors declare no conflict of interest.

Data Availability Statement

The data and code with the analyses supporting our findings are available at https://github.com/ezattara/patagonian_salmonid_distribution

References

Aigo J., Cussac V., Peris S., Ortubay S., Gómez S., López H., et al.(2008) Distribution of introduced and native fish in Patagonia (Argentina): patterns and changes in fish assemblages. Reviews in Fish Biology and Fisheries18, 387–408.

Angermeier P.L. & Winston M.R. (1999) Characterizing fish community diversity across virginia landscapes: prerequisite for conservation. Ecological Applications9, 335–349.

Arismendi I., Doris S., Penaluna B., Jara C., Leal C. & León-Muñoz J. (2009) Aquaculture, non-native salmonid invasions and associated declines of native fishes in Northern Patagonian lakes. Freshwater Biology54, 1135–1147.

Arismendi I., Penaluna B.E., Dunham J.B., García de Leaniz C., Soto D., Fleming I.A., et al.(2014) Differential invasion success of salmonids in southern Chile: Patterns and hypotheses. Reviews in Fish Biology and Fisheries24, 919–941.

Arismendi, I., Penaluna, B., Gomez-Uchida, D., Di Prinzio, C., Rodríguez-Olarte, D., Carvajal-Vallejos, F. M., & Savaria, P. (2019). Trout and char of South America. Trout and char of the world, 279-311.

Baigún C. & Quirós R. (1985) Introducción de Peces exóticos en la República Argentina. Mar del Plata.

Bain M.B. & Stevenson N.J. (1999) Aquatic Habitat Assessment:common methods. American Fisheries Society, Bethesda.

Barriga J.P., Espinós N. a., Chiarello-Sosa J.M. & Battini M. a. (2013) The importance of substrate size and interstitial space in the microhabitat selection by the stream-dwelling catfish Hatcheria macraei (Actinopterygii, Trichomycteridae). Hydrobiologia705, 191–206.

Barros V.., Cordon V.., Moyano C.., Mendez R.., Forquera J.. & Pizzio O. (1983) Cartas de precipitación de la zona oeste de las Provincias de Rio Negro y Neuquén: Primera contribución.Buenos Aires. Belyea L.R. & Lancaster J. (1999) Assembly Rules within a Contingent Ecology. Oikos86, 402.

Booth B.D. & Swanton C.J. (2006) Assembly theory applied to weed communities. Weed Science.

Bozek M.A. & Hubert W.A. (1992) Segregation of resident trout in streams as predicted by three habitat dimensions. Canadian Journal of Zoology70, 886–890.

Cadotte M.W. & Tucker C.M. (2017) Should Environmental Filtering be Abandoned? Trends in Ecology and Evolution32, 429–437.

Coleman M.A. & Fausch K.D. (2007) Cold Summer Temperature Limits Recruitment of Age-0 Cutthroat Trout in High-Elevation Colorado Streams. Transactions of the American Fisheries Society.

Creque S.M., Rutherford E.S. & Zorn T.G. (2005) Use of GIS-Derived Landscape-Scale Habitat Features to Explain Spatial Patterns of Fish Density in Michigan Rivers. North American Journal of Fisheries Management25, 1411–1425.

Cussac V., Becker L., Aigo J., Conte-Grand C., Blasetti G., Cordero P., et al. (2014) Abundance of native fishes, wild-introduced salmonids and escaped farmed rainbow trout in a Patagonian reservoir. Lakes and Reservoirs: Research and Management19, 74–85.

Cussac V.E., Ortubay S., Iglesias G., Milano D., Lattuca E., Barriga J.P., et al. (2004) The distribution of South American galaxiid fishes : the role of biological traits and post-glacial history. Journal of Biogeography, 103–121.

E. Habit, J. González, J. Ortiz-Sandoval, A. Elgueta C.S. (2015) Efectos de la invasión de salmónidos en ríos y lagos de Chile. Ecosistemas24, 43–51.

Fausch K.D. (2008) A paradox of trout invasions in North America. Biological Invasions10, 685–701.

Fausch K.D. (1989) Do gradient and temperature affect distribution of, and interactions between, brook charr (Salvelinus fontinalis) and other resident salmonids in stream? Physiology and Ecology, Japan, 303–322.

Fausch K.D. (2007) Introduction, establishment and effects of non-native salmonids: Considering the risk of rainbow trout invasion in the United Kingdom. Journal of Fish Biology71, 1–32.

Fausch K.D., Baxter C. V. & Murakami M. (2010) Multiple stressors in north temperate streams: Lessons from linked forest-stream ecosystems in northern Japan. Freshwater Biology.

Fausch K.D., Taniguchi Y., Nakano S., Grossman G.D. & Townsend C.R. (2001) Flood disturbance regimes influence rainbow trout invasion success among five holartic regions. Ecological Applications11, 1438–1455.

Fischer J.R. & Paukert C.P. (2008) Habitat relationships with fish assemblages in minimally disturbed Great Plains regions. Ecology of Freshwater Fish17, 597–609.

Frimpong E.A., Sutton T.M., Engel B.A. & Simon T.P. (2005) Spatial-scale effects on relative importance of physical habitat predictors of stream health. Environmental Management.

Frissell C. a, Liss W.J., Warren C.E. & Hurley M.D. (1986) A hierarchical framework for stream habitat classification: viewing streams in a watershead context. Environmental Management10, 199–214.

Gido K.B., Falke J.A., Oakes R.M. & Hase K.. (2006) Fish-habitat relations across spatial scales in prairie streams. American Fisheries Society Symposium48, 265–285.

Grossman G.D., Warnell D.B. & Sabo J.L. (2010) Incorporating Environmental Variation into Models of Community Stability: Examples from Stream Fish.

Habit E., Gonzalez J., Ruzzante D.E. & Walde S.J. (2012) Native and introduced fish species richness in Chilean Patagonian lakes: Inferences on invasion mechanisms using salmonid-free lakes. Diversity and Distributions18, 1153–1165.

E Habit, J González, J Ortiz-Sandoval, A Elgueta, C Sobenes.2015. Efectos de la invasión de salmónidos en ríos y lagos de Chile. Ecosistemas 24 (1), 43-51

Harig A.L. & Fausch K.D. (2002) Minimum habitat requirements for establishing translocated cutthroat trout populations. Ecological Applications.

Heggenes J. & Traaen T. (1988) Downstream migration and critical water velocities in stream channels for fry of four salmonid species. Journal of Fish Biology32, 717–727.

Hubert N. & Renno J.-F. (2006) Historical biogeography of South American freshwater fishes. Journal of Bio, 1414–1436.

Imhol J.G., Fitzgibbon J. & Annable W.K. (1996) A hierarchical evaluation system for characterizing watershed ecosystems for fish habitat. Canadian Journal of Fisheries and Aquatic Sciences53, 312–326.

Juncos R., Beauchamp D.A. & Vigliano P.H. (2013) Modeling prey consumption by native and nonnative piscivorous fishes: Implications for competition and impacts on shared prey in an ultraoligotrophic lake in patagonia. Transactions of the American Fisheries Society142, 268–281.

Karr J.R., Yant P.R., Fausch K.D. & Schlosser I.J. (1987) Spatial and Temporal Variability of the Index of Biotic Integrity in Three Midwestern Streams. Transactions of the American Fisheries Society116, 1–11.

Kolar, C. S., & Lodge, D. M. (2001). Progress in invasion biology: predicting invaders.Trends in ecology & evolution,16(4), 199-204

Korsu K., Huusko A. & Muotka T. (2008) Ecology of alien species with special reference to stream salmonids. Boreal Environment Research13, 43–52.

Kraft N.J.B., Adler P.B., Godoy O., James E.C., Fuller S. & Levine J.M. (2015) Community assembly, coexistence and the environmental filtering metaphor. Functional Ecology29, 592–599.

Lallement M.E. (2017) Plantilla de hábitat y peces de arroyos tributarios del lago Nahuel Huapi , Patagonia. Universidad Nacional del Comahue.

Macchi P.J., Cussac V.E., Alonso M.F., Denegri M.A. & Macchi V.E. Cussac, M.F. Alonso, y M.A. Denegri P.J. (1999) Predation relationships between introduced salmonids and the native fish fauna in lakes and reservoirs in norther Patagonia. Ecology of Freshwater Fish8, 227–236.

Macchi P.J. & Vigliano P.H. (2014) Salmonid introduction in Patagonia: the ghost of past, present and future management. Ecología Austral24, 162–172.

Marchetti M.P., Light T., Moyle P.B. & Viers J.H. (2011) Fish Invasions in California Watersheds : Testing Hypotheses Using Landscape Patterns 14, 1507–1525.

Marchetti M.P., Moyle P.B. & Levine R. (2004) Alien fishes in California watersheds: Characteristics of successful and failed invaders. Ecological Applications14, 587–596.

Matthews, W.J., (1998). Patterns in Freshwater Fish Ecology. Chapman & Hall, NewYork, New York.

McDowall R.M. (2003) Impacts of Introduced Salmonids on Native Galaxiids in New Zealand Upland Streams: A New Look at an Old Problem. Transactions of the American Fisheries Society132, 229–238.

Miserendino M.L. (2007) Macroinvertebrate functional organization and water quality in a large arid river from Patagonia (Argentina). Annales de Limnologie - International Journal of Limnology43, 133–145.

Modenutti B., Albariño R., Navarro M.B., Villanueva V.D., Sol Souza M., Trochine C., et al. (2010) Structure and dynamic of food webs in Andean North Patagonian freshwater systems: Organic matter, light and nutrient relationships. Ecologia Austral20, 95–114.

Montgomery D.R., Beamer E.M., Pess G.R. & Quinn T.P. (1999) Channel type and salmonid spawning distribution and abundance. Canadian Journal of Fisheries and Aquatic Sciences56, 377–387.

Moyle P.B. & Light T. (1996) Biological invasions of fresh water: empirial rules and assembly theory. Biological Conservation78, 149–161.

Nehring R.B. & Anderson R.M. (1993) Determination of population-limiting critical salmonid habitats in Colorado streams using the physical habitat simulation system. Rivers4, 1–19.

Oakes R.M., Gido K.B., Falke J.A., Olden J.D. & Brock B.L. (2005) Modelling of stream fishes in the Great Plains, USA. Ecology of Freshwater Fish14, 361–374.

Paruelo J.M., Beltrán a, Jobbágy E., Sala O.E. & Golluscio R. (1998) The climate of Patagonia: gneral patterns and control son biotic processes. Ecología Austral8, 85–102.

Pascual M., Pascual M., Macchi P., Macchi P., Urbanski J., Urbanski J., et al.(2002) Evaluating potential effects of exotic freshwater fish from imcomplete species presence-absence data. Biological Invasions4, 101–113.

Pascual M.A., Cussac V., Dyer B., Soto D., Vigliano P., Ortubay S., et al. (2005) Freshwater Fishes , Their Biodiversity , Fisheries and Habitats: Health Chapter Title: Freshwater fishes of Patagonia in the 21st Century after a hundred years of human settlement , species introductions , and environmental change. Water Resources.

Pascual M.A., Cussac V., Dyer B., Soto D., Vigliano P., Ortubay S., et al. (2007) Freshwater fishes of Patagonia in the 21st Century after a hundred years of human settlement, species introductions, and environmental change. Aquatic Ecosystem Health & Management10, 212–227.

Rahel F.J. & Nibbelink N.P. (1999) Spatial patterns in relations among brown trout ( Salmo trutta ) distribution, summer air temperature, and stream size in Rocky Mountain streams. Canadian Journal of Fisheries and Aquatic Sciences56, 43–51.

Reeves G.H., Bisson P. & DambacherJ.M (1998) Fish communities. In: River ecology and management: lessons from the Pacific coastal ecoregion. (Eds R.J. Naiman & R.E. Bilby), pp. 200–234. Srpinger-Verlag, New York. Rowe D.C., Pierce C.L. & Wilton T.F. (2009) Natural Resource Ecology and Management Publications Natural Resource Ecology and Management.

Ruzzante D.E., Walde S.J., Cussac V.E., Dalebout M.L., Seibert J., Ortubay S., et al. (2006) Phylogeography of the Percichthyidae (Pisces) in Patagonia: Roles of orogeny, glaciation, and volcanism. Molecular Ecology15, 2949–2968.

Schlosser I.J. & Kallemeyn L.W. (2000) Spatial Variation in Fish Assemblages across a Beaver-Influenced Successional Landscape. Ecology81, 1371.

Smith T. a. & Kraft C.E. (2005) Stream Fish Assemblages in Relation to Landscape Position and Local Habitat Variables. Transactions of the American Fisheries Society134, 430–440.

Soto D., Arismendi I., González J., Sanzana J., Jara F., Jara C., et al. (2006) Southern Chile, trout and salmon country: invasion patterns and threats for native species. Revista chilena de historia natural79, 97–117.

Southwood T.R.E. (1977) Habitat, the Templet for Ecological Strategies? The Journal of Animal Ecology46, 336.

Stanfield L.W., Scott F. & Borwick J.A. (2006) Using a landscape approach to identify the distribution and density patterns of salmonids in Lake Ontario tributaries. American Fisheries Society Symposium 4848, 601–621.

Strange E.M. & Foin T.C. (2009) Interaction of physical and biological processes in the assembly of stream fish communities. In: Ecological Assembly Rules. .

Vannote R.L., Minshall G.W., Cummins K.W., Sedell J.R. & Cushing C.E. (1980) The river continuum concept. Canadian Journal of Fisheries and Aquatic Sciences37, 130–137.

Warren M., Dunbar M.J. & Smith C. (2015) River flow as a determinant of salmonid distribution and abundance: a review. Environmental Biology of Fishes98, 1695–1717.

Weiher, E., & Keddy, P. A. (1995). The Assembly of Experimental Wetland Plant Communities.Oikos,73(3), 323–335.

Session Information

sessionInfo()

## R version 4.0.2 (2020-06-22)
## Platform: x86_64-w64-mingw32/x64 (64-bit)
## Running under: Windows 10 x64 (build 19041)
## 
## Matrix products: default
## 
## locale:
## [1] LC_COLLATE=English_United States.1252 
## [2] LC_CTYPE=English_United States.1252   
## [3] LC_MONETARY=English_United States.1252
## [4] LC_NUMERIC=C                          
## [5] LC_TIME=English_United States.1252    
## 
## attached base packages:
## [1] grid      stats     graphics  grDevices utils     datasets  methods  
## [8] base     
## 
## other attached packages:
##  [1] factoextra_1.0.7    Hmisc_4.5-0         Formula_1.2-4      
##  [4] survival_3.2-12     lattice_0.20-44     VennDiagram_1.6.20 
##  [7] futile.logger_1.4.3 gt_0.3.1            forcats_0.5.1      
## [10] stringr_1.4.0       dplyr_1.0.7         purrr_0.3.4        
## [13] readr_2.0.1         tidyr_1.1.3         tibble_3.1.3       
## [16] ggplot2_3.3.5       tidyverse_1.3.1    
## 
## loaded via a namespace (and not attached):
##  [1] fs_1.5.0             bit64_4.0.5          lubridate_1.7.10    
##  [4] RColorBrewer_1.1-2   httr_1.4.2           tools_4.0.2         
##  [7] backports_1.2.1      bslib_0.2.5.1        utf8_1.2.2          
## [10] R6_2.5.0             rpart_4.1-15         DBI_1.1.1           
## [13] colorspace_2.0-2     nnet_7.3-16          withr_2.4.2         
## [16] tidyselect_1.1.1     gridExtra_2.3        curl_4.3.2          
## [19] bit_4.0.4            compiler_4.0.2       cli_3.0.1           
## [22] rvest_1.0.1          formatR_1.11         htmlTable_2.2.1     
## [25] xml2_1.3.2           labeling_0.4.2       sass_0.4.0          
## [28] scales_1.1.1         checkmate_2.0.0      digest_0.6.27       
## [31] foreign_0.8-81       rmarkdown_2.10       rio_0.5.27          
## [34] base64enc_0.1-4      jpeg_0.1-9           pkgconfig_2.0.3     
## [37] htmltools_0.5.1.1    highr_0.9            dbplyr_2.1.1        
## [40] htmlwidgets_1.5.3    rlang_0.4.11         readxl_1.3.1        
## [43] rstudioapi_0.13      farver_2.1.0         jquerylib_0.1.4     
## [46] generics_0.1.0       jsonlite_1.7.2       vroom_1.5.4         
## [49] zip_2.2.0            car_3.0-11           magrittr_2.0.1      
## [52] Matrix_1.3-4         Rcpp_1.0.7           munsell_0.5.0       
## [55] fansi_0.5.0          abind_1.4-7          lifecycle_1.0.0     
## [58] stringi_1.7.3        yaml_2.2.1           carData_3.0-4       
## [61] parallel_4.0.2       ggrepel_0.9.1        crayon_1.4.1        
## [64] haven_2.4.3          splines_4.0.2        hms_1.1.0           
## [67] knitr_1.33           pillar_1.6.2         ggpubr_0.4.0        
## [70] ggsignif_0.6.2       futile.options_1.0.1 reprex_2.0.1        
## [73] glue_1.4.2           evaluate_0.14.1      latticeExtra_0.6-29 
## [76] lambda.r_1.2.4       data.table_1.14.0    modelr_0.1.8        
## [79] png_0.1-8            vctrs_0.3.8          tzdb_0.1.2          
## [82] cellranger_1.1.0     gtable_0.3.0         assertthat_0.2.1    
## [85] openxlsx_4.2.4       xfun_0.25            broom_0.7.9         
## [88] rstatix_0.7.0        cluster_2.1.2        ellipsis_0.3.2